Alibaba

ByteDance

CATL

Huawei

Meituan

Tencent

Xiaomi

Most Followed

Market News

Listings (IPO)

Earnings Reports

Venture Roundup

Big Tech

Automotives

Retail and Commerce

Collections

News

Features

Insights

Press Releases

Anta

DeepSeek

Geely

Li Auto

Mixue Bingcheng

Pop Mart

Going Global

Latest

Collaborate

Earnings

Auto

Retail & Commerce

Multiple independent sources told 36Kr that Z.ai’s annual recurring revenue (ARR) reached USD 1 billion in July.
Over the past year, coding and video generation models have emerged as two of the <a href="/industry/ai">artificial intelligence</a> sector’s strongest revenue drivers.
Outside China, Anthropic’s Claude Code reached USD 1 billion in ARR just six months after its launch. The product’s growth also helped lift Anthropic’s valuation to nearly USD 1 trillion in May, surpassing OpenAI.
Investors, however, had long questioned whether Chinese foundation model companies could replicate Anthropic’s financial success.
Anthropic benefits from operating in North America, where businesses are more willing to pay for software and the subscription ecosystem is more mature. Its valuation is also roughly an order of magnitude above Z.ai’s latest market capitalization, giving it access to a deeper pool of capital for the computing capacity and other resources needed to train models. Chinese developers have yet to match that advantage.
But an investor told 36Kr that Z.ai reached the milestone much sooner than expected. The company had not been projected to reach an ARR of USD 1.0–1.5 billion until the end of 2026, the investor said.
A direct comparison illustrates the pace of its growth. According to Anthropic’s disclosures and Z.ai’s financial reports, Anthropic took 15 months to increase its ARR from USD 100 million to USD 1 billion. Z.ai took five months.
A person familiar with the matter said Z.ai’s ARR grew 15-fold from January to July this year.
Z.ai was among the first Chinese artificial intelligence model developers to focus heavily on coding as a use case. In early 2025, it began concentrating its resources on improving its models’ coding capabilities. Founder Jie Tang has described coding and reasoning as central to Z.ai’s strategy, arguing that both capabilities can advance alongside AI agents and support their development.
Z.ai subsequently released a series of flagship models, averaging about one every two months. Their coding performance placed them among the leading models globally.
The company released its latest model, GLM-5.2, last month. The open-source model reportedly matched or outperformed Claude Opus 4.8 and GPT-5.5 on several benchmarks.
Z.ai has said the commercial value of foundation models depends on the relationship between model intelligence and token consumption. The company argues that usage will continue to grow as long as a model maintains a capability lead.
Z.ai’s financial results showed that it raised GLM API prices by a cumulative 83% in the first quarter. Its international subscription prices approached those of Claude Code, while usage increased by about 400%.
As token consumption becomes an increasingly important measure of model monetization, coding and video generation models represent two distinct business models. Coding models power productivity tools sold directly to businesses and developers, while video-generation models are increasingly being integrated into industrial-scale content production workflows.
Revenue growth suggests that demand for coding models is becoming more established. A similar pattern is emerging in video generation. Seedance 2.0 is reportedly generating nearly RMB 1 billion (USD 147.5 million) in monthly revenue. Goldman Sachs, meanwhile, forecasts that Kling AI’s ARR could reach USD 1 billion by the end of the year.
As the commercial opportunity becomes clearer, competition is intensifying.
MiniMax released M3 in June, with a focus on strengthening its coding and agent capabilities. On July 16, Moonshot AI released Kimi K3, an open-source model with 2.8 trillion parameters. Its overall intelligence performance reportedly trails only Claude Fable 5 and GPT-5.6 Sol.
Competition is also intensifying outside China. By releasing GPT-5.6 and integrating ChatGPT with Codex, OpenAI is seeking to regain ground in AI coding.
KrASIA features translated and adapted content that was originally published by 36Kr. This <a href="https://36kr.com/p/3898662052693894">article</a> was written by Zhou Xinyu for 36Kr.
Note: RMB figures are converted to USD at rates of RMB 6.78 = USD 1 based on estimates as of July 17, 2026, unless otherwise stated. USD conversions are presented for ease of reference and may not fully match prevailing exchange rates.

China’s Z.ai reaches USD 1 billion in ARR after 15-fold growth in six months

Artificial Intelligence

Deep Tech

Foundation and Large Models

China

… Do you agree that AI has brought all players back to the same starting line?</h3>
WX: Not really.
Scale effects are very important. To be honest, for the same thing, investing 800 people and investing only two or three people are completely different. Behind a product, there is more engineering than algorithms.
Engineering requires real investment. It cannot be solved just by vibe coding and being smart enough. That is why we built shared product platforms, trying to turn the dirty and labor-intensive parts of engineering into reusable pipelines.
<h3>36Kr: Meitu is a company that has succeeded many times, and each of its products is in a highly competitive space. What have you learned?</h3>
WX: Most successful products begin in scenarios where there is no competition. For the four new products released this time, there are hardly any competitors in the market. We are only competing with ourselves.
Also, Meitu is best at running marathons. We have competed with <a href="/company/tencent">Tencent's</a> Pitu and <a href="/company/baidu">Baidu</a> Motu many times before. If you persist, product quality will speak for itself.
KrASIA features translated and adapted content that was originally published by 36Kr. This <a href="https://36kr.com/p/3877112973733895">article</a> was written by Zhou Xinyu for 36Kr.
Note: RMB figures are converted to USD at rates of RMB 6.78 = USD 1 based on estimates as of July 16, 2026, unless otherwise stated. USD conversions are presented for ease of reference and may not fully match prevailing exchange rates.

For Meitu, the next AI product may come from data, passion, or both

… home, you first need to ask what it is entering the home to do. If it is entering to do work, meaning housework, that is a world-class problem at this moment. We are exploring it, but we will not bring it to market so quickly.
But if it is entering the home to provide emotional value plus partial functional value, such as addressing anxiety in the AI era, English learning, and companionship, then current technology already exceeds scenario demand. So the bottom line for entering the home is safety, and the upper bound is intelligence. If we hold the bottom line and continue exploring upward at the upper bound, household scenarios will gradually open up, moving step by step from simple tasks and simple scenarios toward complexity.
Behind each step, we have a complete system plan for the body, low-level motion-control system, and AI brain.
<h3>36Kr: Some in the industry say development of the robot body has hit a bottleneck, and intelligence is the next challenge. What do you think?</h3>
TH: This depends on the scenario. Intelligence is indeed significant, and it is used to solve work and create productivity value. But body hardware is far from truly mature. Today’s bodies have, at most, reached what is considered usable. They are still quite far from being easy to use. This requires technology advancement in robot joints, sensors, lightweight materials, battery energy density, and more. There is still a long way to go.
<h3>36Kr: Standardization in the robotics supply chain seems poor, and each manufacturer tends to be fairly nonstandardized. Will the industry move toward full standardization and large-scale mass production like the automotive sector?</h3>
TH: China produces more than 30 million cars a year, all standardized, with one rolling off the production line every minute. Robots will certainly reach a high level of standardization in the future as well.
Right now, robotics is in the middle of moving from a nonstandardized state to a standardized one. We are also calling for this and actively participating. At the national level, work has already begun to organize committees and other bodies to formulate industry standards for embodied intelligence. Clear standards for core components are worth looking forward to.
KrASIA features translated and adapted content that was originally published by 36Kr. This <a href="https://36kr.com/p/3877003233898501">article</a> was written by Peng Li for 36Kr.

Through Qiyuan, Swancor wants to make personal robots widely affordable

Robotics

Chinese retailer <a href="/company/miniso">Miniso</a> is betting its next phase of overseas growth on a portfolio of proprietary IP, backed by plans to open 100 stores in the US this year.
The company has signed 30 artists globally, including those from Japan, South Korea, and Thailand, prioritizing original characters rather than licensed external brands for future growth, Miniso CMO Robin Liu told Nikkei Asia. About a third of those artists have had their characters turned into merchandise that is now on sale.
The New York Stock Exchange-listed company has been collaborating with brands like Disney's Marvel, Peanuts, and Japan's Sanrio since 2019 to produce lifestyle and apparel products using its network of suppliers. Miniso now wants to build character branding it owns outright.
"We need the capability to incubate and operate these IPs, create higher commercial value," Liu said in an interview. "This also ensures our product offering is unique."
The Chinese brand, which opened its first store in Guangzhou in 2013, operates thousands of discount variety shops selling goods that range from stationery and toys to cosmetics and clothing.
Its first original character, "Yoyo," launched in June last year and within six months garnered RMB 100 million (USD 14.7 million) in sales, primarily in China. Liu expects the character to generate roughly RMB 800 million (USD 117.8 million) in revenue this year.
Liu said the company visited Netflix and Sony Pictures to explore the possibilities of turning Miniso's characters into a film or other content.
"We might engage in deeper collaborations in the future, but I think it will be a gradual process," he added.
To introduce Miniso's characters to the US market and raise its brand awareness, the company launched an exhibition at New York's Grand Central Station, featuring 50 Yoyo character sculptures last month. It opened a similar gallery in Shanghai's Bund City Hall Plaza in May.
<figure id="attachment_163015" aria-describedby="caption-attachment-163015" style="width: 1920px" class="wp-caption alignnone"><img loading="lazy" decoding="async" class="size-full wp-image-163015" src="https://console.0722vps.aieehr01.top/wp-content/uploads/2026/07/Miniso-Yoyo-Exhibition-New-York.jpg" alt="Photo shows Miniso's Yoyo character on display at a New York exhibition." width="1920" height="1280" srcset="https://console.0722vps.aieehr01.top/wp-content/uploads/2026/07/Miniso-Yoyo-Exhibition-New-York.jpg 1920w, https://console.0722vps.aieehr01.top/wp-content/uploads/2026/07/Miniso-Yoyo-Exhibition-New-York-720x480.jpg 720w, https://console.0722vps.aieehr01.top/wp-content/uploads/2026/07/Miniso-Yoyo-Exhibition-New-York-1024x683.jpg 1024w, https://console.0722vps.aieehr01.top/wp-content/uploads/2026/07/Miniso-Yoyo-Exhibition-New-York-768x512.jpg 768w, https://console.0722vps.aieehr01.top/wp-content/uploads/2026/07/Miniso-Yoyo-Exhibition-New-York-1536x1024.jpg 1536w" sizes="(max-width: 1920px) 100vw, 1920px" /><figcaption id="caption-attachment-163015" class="wp-caption-text">Miniso's Yoyo character, on display at a New York exhibition. Photo courtesy of Miniso.</figcaption></figure>
The company will use "immersive retail" to showcase the growing portfolio of characters, a concept that has boosted sales in China, according to Tom Bartlebaugh, CEO of Miniso USA.
Miniso plans to open the first three experiential stores in Las Vegas, Columbus, Ohio, and East Rutherford, New Jersey, featuring interactive attractions such as claw machines. Existing US stores will also be remodeled.
Miniso, which entered the US in 2017, now has 380 stores and plans to expand its footprint by 20%. Building brand recognition in the US remains a challenge, Bartlebaugh said. The company will increasingly target strip malls and larger street-front locations rather than traditional shopping malls.
The US is the top priority because it is the company's largest revenue market outside of China, Liu said.
The focus also reflects a tougher retail environment in China, where a <a title="" href="https://asia.nikkei.com/spotlight/policy-asia/china-s-consumption-slump-shrouds-next-5-year-plan" target="_blank" rel="noopener">consumer spending slowdown</a> remains subdued after a yearslong property slump. Despite that, the company has continued to grow domestically and believes its expanding roster of characters will help drive further growth.
Miniso's shift toward a character-based portfolio doubles up with its subsidiary brand Top Toy, which sells blind boxes with figurines and plush toys, analysts noted.
Liu said the two brands will complement each other and can share marketing resources. But Top Toy is focused on being designer art toys, it is more professional and high end than Miniso's mass market appeal, he added.
Loo Wee Teck, global insight manager for toys and games at Euromonitor International, said there is demand for licensed toys but customers are drawn to well-known franchises.
He added many characters are released into the market, but few actually become as widespread as Chinese toymaker <a href="/company/pop-mart">Pop Mart's</a> Labubu.
"Miniso can compete in distribution and price, but building a character with lasting global resonance is a much harder challenge," said Loo.
Lina Yan, HSBC's consumer analyst, said it took Pop Mart nearly two decades to scale one of their other characters to revenue levels of RMB 3 billion (USD 441.9 million) last year.
"Miniso could still grow meaningfully in this segment if it can successfully build and compound a dedicated fan base—rather than simply selling in-house IP toys to its existing customers by substituting for other product categories," she said.
Liu said Miniso views Pop Mart as the industry's benchmark and is studying its approach to developing IP. Together, Miniso and Top Toy aim to sign on and manage 100 artists by year's end.
"The market can't be dominated by just one company like Pop Mart," Liu said.
This article first appeared on <a href="https://asia.nikkei.com/business/consumer/miniso-banks-on-finding-the-next-labubu-as-it-opens-100-us-stores">Nikkei Asia</a>. It has been republished here as part of 36Kr’s ongoing partnership with <a href="https://www.0722vps.aieehr01.top/nikkei-teams-up-with-36kr-in-asia-tech-news-coverage/">Nikkei</a>.
Note: RMB figures are converted to USD at rates of RMB 6.79 = USD 1 based on estimates as of July 14, 2026, unless otherwise stated. USD conversions are presented for ease of reference and may not fully match prevailing exchange rates.

Miniso banks on finding the next Labubu as it opens 100 US stores

Miniso

East Asia

Retail

Consumer Economy

… technical service teams, increasing standardized warehousing capacity, optimizing local inventory turnover management, shortening spare parts delivery times, and strengthening site-specific service capabilities for large vehicle and chemical plants.
Beyond Thailand and Indonesia, ZKH plans to expand into markets such as Vietnam and Malaysia through an asset-light model. By leveraging its existing regional warehouse network, cross-border logistics capabilities, and supplier resources, the company aims to build a more connected MRO service network across Southeast Asia.
ZKH will continue focusing on four key industries: automotive, electronics, chemicals, and equipment manufacturing, providing integrated MRO procurement solutions covering industrial consumables and automation-related spare parts. Through its innovation and R&amp;D center in Taicang, Suzhou, China, the company will further develop overseas offerings of private-label industrial products, including automation components, PPE, labor protection supplies, and industrial lubricants.
It will also continue iterating its cross-border digital material management system to unify material standards across countries and reduce the labor costs companies face in managing materials across regions.
<h3>Setting a benchmark for Southeast Asia’s MRO supply chain</h3>
While serving Chinese-invested factories, ZKH Industrial Supply is also promoting standardized upgrades among local suppliers in Southeast Asia, improving the region’s industrial support system, and enabling more coordinated development across Chinese and overseas industrial chains.
In 2026, ZKH will continue deepening its localized presence in Southeast Asia and improving its regional warehouse network and local supply chain system. Through digital technology, it aims to drive global supply chain coordination, help more Chinese manufacturers establish stable operations in Southeast Asia, address overseas MRO procurement pain points, and support the globalization of Chinese manufacturing.
<h3>About ZKH Industrial Supply</h3>
ZKH Industrial Supply is a global provider of digital MRO procurement and industrial supply chain services. Leveraging a digital procurement platform, extensive product portfolio, and integrated warehousing network, the company supports manufacturers across multiple industries. In Southeast Asia, ZKH operates localized entities in Thailand and Indonesia, providing sourcing, fulfillment, digital material management, and compliance support for manufacturers expanding overseas.
This press release was published in partnership with <a href="https://zkh.global/">ZKH Industrial Supply</a>.

ZKH Industrial Supply targets MRO bottlenecks facing Chinese manufacturers in Southeast Asia

Logistics

On July 10, China's Nexchip debuted on the Hong Kong Stock Exchange after growing into the world's eighth largest semiconductor foundry on its strength in older-generation chips.
Supplies of mature-node semiconductors are becoming tight as top foundries focus on advanced chips for <a href="/industry/ai">artificial intelligence</a>. That trend is a tailwind for latecoming Chinese players like Nexchip.
Nexchip raised HKD 6.9 billion (USD 880.3 million) through its IPO and related share issuance. Its stock opened 11% above the offering price.
Nexchip was established in 2015 under the leadership of a local government in China. Funding came primarily from a state-backed investment fund, while Taiwan's Powerchip also invested and provided technical expertise.
Powerchip's former president, Tsai Kuo-chih, serves as Nexchip's chairman. Powerchip-affiliated investors now hold just over 7% of the company, but their ties still run deep.
At the listing ceremony in Hong Kong, Tsai said the company would increase R&amp;D investment and production capacity, while speeding up its global expansion. He said the IPO shows the confidence that international capital markets have in China's semiconductor sector.
Nexchip listed on the Star Market, the Shanghai Stock Exchange's technology-focused board, in 2023. The dual listing in Hong Kong is aimed at reaching out to international investors and enables the company to raise Hong Kong dollars, which are more readily convertible into foreign currency, facilitating overseas purchases of equipment and technology.
Nexchip produces chips used in display panels, as well as image sensors, power management semiconductors, and other products. Its most advanced manufacturing process used in mass production is the 40-nanometer node, a technology that has been in use for nearly two decades.
The company's growth picked up thanks to the semiconductor shortage triggered by the Covid-19 pandemic. Supported by government policies, Nexchip continued expanding production even after market conditions softened. Its annual output reached 1.67 million 300-millimeter wafer equivalents in 2025, up 74% from 2023.
Nexchip ranked eighth worldwide in semiconductor foundry revenue in this year's January-March quarter, according to Taiwan-based research firm TrendForce. It surpassed Taiwan's Powerchip Semiconductor Manufacturing Corporation, the company that gave it its technological start.
Chip market conditions remain favorable. As industry leaders such as Taiwan Semiconductor Manufacturing Company concentrate on advanced AI chips, TrendForce said there are signs that prices for several generations of technologies could rise by 5–10% during the July-September quarter.
Taiwan's United Microelectronics Corporation, which also produces mature-node chips, reported that June revenue rose 22.9% on the year.
China's largest foundry, Semiconductor Manufacturing International Corporation (SMIC), said in a May earnings briefing that semiconductor production capacity outside China is increasingly being allocated to AI-related products.
"Consumer electronics makers and other customers are seeking to secure manufacturing capacity [for chips] in China, and orders are returning," SMIC co-CEO Zhao Haijun said.
Typically, late entrants in mature-node semiconductor manufacturing face fierce competition. Building fabrication facilities requires big investments, and established players that have already depreciated their production assets enjoy significant cost advantages. Meanwhile, customers move on to more advanced chips over time.
But Chinese chipmakers have continued investing heavily despite being latecomers. Capital spending by Chinese foundries from 2015 to 2023 exceeded their revenue by 12%, according to the US-based Semiconductor Industry Association. Companies outside of China invested an amount equivalent to roughly 33% of revenue.
A report by Rhodium Group, a US think tank, noted that Chinese foundries are able to invest under financial constraints that differ from those faced by foreign competitors, thanks in part to government support.
This article first appeared on <a href="https://asia.nikkei.com/business/tech/semiconductors/china-s-nexchip-has-global-ambitions-as-legacy-chips-thrive-in-ai-s-shadow">Nikkei Asia</a>. It has been republished here as part of 36Kr’s ongoing partnership with <a href="https://www.0722vps.aieehr01.top/nikkei-teams-up-with-36kr-in-asia-tech-news-coverage/">Nikkei</a>.
Note: HKD figures are converted to USD at rates of HKD 7.84 = USD 1 based on estimates as of July 14, 2026, unless otherwise stated. USD conversions are presented for ease of reference and may not fully match prevailing exchange rates.

China's Nexchip has global ambitions as legacy chips thrive in AI's shadow

Hong Kong

… Shenzhen-based AI hardware company, has raised an eight-figure RMB sum in an angel round. Investor details were not disclosed. The company is developing home AI infrastructure comprising voice-enabled routers, network-attached storage devices, and computing boxes that combine connectivity, storage, and local processing. It will use the funds for R&amp;D, production, global channel development, and ecosystem expansion. —<a href="https://36kr.com/p/3879780282495236">36Kr</a></li>
<li aria-level="1">ZeroErr, a Shenzhen-based developer of high-precision robotic joints and other core components, has raised a nine-figure RMB sum in a Series C++ round. Cowin Capital led the round, with participation from Guotai Junan Innovation Investment and existing investor Tsinghua Holdings Capital. The company will use the funds to expand its production capacity and global market presence. —<a href="https://36kr.com/p/3885232033378308">36Kr</a></li>
<li aria-level="1">NNB Nutrition, a Nanjing-based developer of dietary supplement ingredients, has raised nearly RMB 200 million (USD 29.5 million) in a Series C+ round. China Life Private Equity Investment led the round, with participation from existing investor MX Capital. The company will use the funds for product R&amp;D and the construction of a new manufacturing facility. —<a href="https://36kr.com/p/3890940898867712">36Kr</a></li>
<li aria-level="1">Tashan Technology, a Beijing-based developer of AI-powered tactile perception technology, has raised a nine-figure RMB sum in a Series B round. Investors include Joyson Electronics, Taiping Innovation, AUX Group, Pengling, Lavender Hill Capital Partners, and Hongshan Innovation Capital. Existing investors Dowstone and BeFor Capital also made follow-on investments. The company will use the funds to upgrade its tactile sensors and chips, deploy application-specific solutions, and build a tactile data and training platform for <a href="/industry/robotics">robotics</a>. —<a href="https://36kr.com/p/3889141540370949">36Kr</a></li>
<li aria-level="1">DeepCtrls, a physical AI company serving the electromechanical and energy infrastructure sectors, has raised a nine-figure RMB sum in a Series B round. SDIC Fund and CMB International jointly led the round, while JinkoSolar made a strategic investment. Existing investors HSG, Source Code Capital, Forebright Capital, and China Merchants Venture also participated. DeepCtrls will use the funds for R&amp;D, the broader deployment of its standardized DeepBot product, and international expansion. —<a href="https://36kr.com/p/3887726503688968">36Kr</a></li>
<li aria-level="1">Thinkhead M&amp;E, a Shanghai-based manufacturer of high-end machine tools, has raised a nine-figure RMB sum in a new round led by Wuliangye Fund, with participation from Megmeet. The company produces precision gear-grinding machines, grinders, and milling machines for industries including automotive manufacturing, robotics, aerospace, and precision reducers. It will use the funds for product R&amp;D, the commercialization of new products, and expansion into overseas markets. —<a href="https://36kr.com/p/3886400975040520">36Kr</a></li>
<li aria-level="1">Snapmaker, a consumer 3D printer manufacturer, has raised RMB 1 billion (USD 147.3 million) in a round led by Cathay Capital, with TAL Education making a strategic follow-on investment. Existing shareholders <a href="/company/meituan">Meituan</a>, DragonBall Capital, GL Ventures, and Shunwei Capital also increased their commitments. Snapmaker did not disclose how it would allocate the proceeds. Its current priorities include expanding its library of printable models and applying AI tools to make 3D design and creation more accessible. —<a href="https://36kr.com/p/3885728813691145">36Kr</a></li>
<li aria-level="1">MindRank, an AI-native biopharmaceutical company, has raised USD 52 million in a Series B round. Investor details were not disclosed, although the company said the round involved major RMB- and USD-denominated funds. MindRank will use the funds to upgrade its platform, enhance its multi-agent and clinical data feedback systems, advance oral GLP-1 drug candidate MDR-001 through Phase 3 trials and commercialization, and expand its pipeline of differentiated drug candidates. —<a href="https://36kr.com/p/3885479689465858">36Kr</a></li>
<li aria-level="1">Xingji Yuandian, a Shanghai-based provider of space-computing satellite systems, has completed seed and angel rounds totaling an eight-figure RMB sum. Unity Ventures backed the seed round. The angel round was jointly led by Unity Ventures, Plum Ventures, and a fund under Shanghai State-owned Capital Investment, with Shanghai Angel Group participating through a co-investment arrangement. The company did not disclose a detailed allocation of the proceeds. It is developing low-Earth-orbit satellites designed to host computing payloads and support a future orbital data center. —<a href="https://36kr.com/p/3885413484163074">36Kr</a></li>
</ul>
whereRelatedArticle
<h3>Amity Robotics, Acti, Qashier, and others made recent headlines:</h3>
<ul>
<li aria-level="1">Amity Robotics, a Thailand-based company, raised USD 7 million in seed funding through a mix of equity and debt financing. East Ventures led the equity round, with participation from 500 Global, while AlteriQ Global led the debt facility.</li>
<li aria-level="1">Acti, a Singapore-based AI startup founded by former <a href="/company/baidu">Baidu</a> executives, raised USD 5.3 million in a seed round led by Bitkraft Ventures.</li>
<li aria-level="1">Qashier, a Singapore-based <a href="/industry/fintech">fintech</a> company, raised USD 6.125 million in a Series A+ round comprising equity and debt financing. The round was led by Cocoon Capital, IFP Securities, and BlackSoil Global, with participation from angel investors.</li>
</ul>
If there are any news or updates you'd like us to feature, get in touch with us at: <a href="mailto:hello@0722vps.aieehr01.top">hello@0722vps.aieehr01.top</a>.
Note: RMB figures are converted to USD at rates of RMB 6.79 = USD 1 based on estimates as of July 14, 2026, unless otherwise stated. USD conversions are presented for ease of reference and may not fully match prevailing exchange rates.

Deals in brief: BlueOrchard backs Malaysia’s PolicyStreet, PixVerse extends Series C round, Ant Group acquires stake in Boohee Health, and more

Ant Group

Singapore

Malaysia

Southeast Asia

When James Peng founded <a href="/company/pony-ai">Pony.ai</a> in 2016, he expected robotaxis, or autonomous taxi services, to need at least a decade of work before moving from long-term vision to large-scale deployment. The technology had to mature. Laws and regulations had to catch up. Public acceptance also had to be tested.
That timeline has largely held. Waymo vehicles now move through the streets of San Francisco, while Pony.ai’s fleet accepts ride-hailing orders in Shenzhen’s busy Nanshan district. <a href="/industry/autonomous-vehicles">Autonomous vehicles</a> have entered public traffic and begun commercial operations.
What Peng may not have anticipated ten years ago was that one of the key constraints on rapid fleet expansion would be a set of routine operations and maintenance tasks that become more complicated once the driver is removed.
When a human is behind the wheel, the driver handles charging, car washing, vehicle maintenance, and even helping passengers with luggage. Once autonomous vehicles enter service, these invisible errands become operational problems of their own.
<a href="https://36kr.com/p/3876767161446665">In an interview</a>, Pony.ai CEO James Peng spoke with 36Kr about robotaxi deployment and operations, the company’s choices around mass production for its Level 2 driver assistance business, its decision not to rush into embodied intelligence, and its technical approach.
Peng told 36Kr that Pony.ai has already built an operations team and developed a set of operations and maintenance standards to support autonomous vehicles. “This workforce includes remote safety operators, as well as ground support and logistics staff,” he said.
Even if the fleet expands rapidly, Peng said the staff-to-vehicle ratio will not rise significantly, breaking from the traditional assumption that more vehicles require more people. He believes this is an operational blind spot that today’s ride-hailing companies and automakers may overlook when entering the robotaxi market.
After nearly a decade of development, driverless vehicle technology has entered a stage of commercial operation across several fields as automotive intelligence has matured across the supply chain. Autonomous logistics vehicles are one example. Rino.ai, Neolix, Zelos, and others have attracted investor backing and become prominent names in China’s logistics sector.
Robotaxis, one of the most visible commercial applications for driverless vehicles, have attracted automakers, ride-hailing platforms, and other companies. Tesla, <a href="/company/xpeng">Xpeng</a>, <a href="/company/geely">Geely</a>, and others have announced plans to operate driverless vehicles.
One mainstream view is that automakers will become key robotaxi players because they have full-vehicle engineering capabilities and driver assistance-to-autonomous driving technology systems that can generate a data flywheel, where more vehicles on the road produce more data to improve the system. But Peng told 36Kr that manufacturing capacity is already abundant in China. For robotaxis, he said, any company that has not operated such a service before is still nearly starting from zero.
Peng believes technical capability answers the zero-to-one question for robotaxis, namely whether a company can build the system at all. Operating capability, by contrast, determines efficiency. Pony.ai is now exploring what it sees as an increasingly clear business model.
The road has not always been smooth. The sector has long faced commercialization challenges, and many autonomous driving companies have shifted their focus toward supplying Level 2 driver assistance systems, which still require human supervision, as they searched for more revenue channels.
Pony.ai is one of the few companies that has remained committed to robotaxis. Its seventh-generation vehicles have already achieved per-vehicle profitability in Guangzhou and Shenzhen. After validating its financial model, Pony.ai plans to expand its fleet to 3,500 vehicles this year.
Looking back at the choices it has made, Pony.ai has often taken a path that runs counter to prevailing assumptions.
Peng said the company briefly tried the large-scale Level 2 driver assistance business, but quickly concluded that it was likely to be low margin because “the technical threshold is low, the user experience is not standardized, automakers hold the bargaining power, and intelligent driving companies can easily fall into price wars.”
On the push by automakers and ride-hailing platforms into robotaxis, Peng was more pointed. “Making an announcement is always easy,” he said. “Tesla has been talking about it for ten years. Has it done it?”
Automakers generally hope to emulate Tesla by using existing end-to-end algorithm capabilities as a common technology base for both driver assistance and robotaxi businesses.
But Peng offered a technical view that differs from the mainstream. Pony.ai, he said, has not followed the trajectory of large language models by building a unified autonomous driving algorithm with a vast number of parameters. Instead, it has continued to use a strategy that combines multiple smaller models, improving operating efficiency and reducing dependence on computing power.
As driver assistance developers move away from maps, Pony.ai has again chosen not to follow the crowd. Peng said the company will continue to use a lightweight mapping approach for the long term.
“You’re familiar with the roads you drive often, so driving them is easier. Places you haven’t been to before are tiring. That’s normal. So why not add maps?” Peng said. “Even Tesla uses maps.”
Pony.ai has also decided against entering embodied intelligence at this stage, choosing instead to watch from the sidelines.
“This is another thing that will take at least ten years,” Peng said directly. “There will always be opportunities to do it, but you still have to see clearly and think clearly.”
KrASIA features translated and adapted content that was originally published by 36Kr. This <a href="https://36kr.com/p/3876767161446665">article</a> was written by 36Kr Auto.

Pony.ai CEO James Peng on the company’s path in autonomous driving

Pony.ai

Transportation

Autonomous Vehicles

Mobility

US companies' use of Chinese <a href="/industry/ai">artificial intelligence</a> services surged in June, when Anthropic suspended some models over government restrictions, as open-source models with prices as low as one-twentieth of Anthropic's Claude Mythos become more entrenched in the American corporate world.
In response to soaring fees, US cryptocurrency exchange Coinbase shifted much of its AI use to cheaper Chinese models as GLM and Kimi. This and other steps have "cut our AI spend nearly in half," CEO Brian Armstrong wrote in a social media post on June 27.
GLM, developed by Z.ai, and Kimi, developed by Moonshot AI, both excel at software coding and time-consuming tasks. Their cost per token is about one-twentieth that of Anthropic's latest models.
GLM-5.2, released in mid-June, is drawing notice for its performance. In testing with roughly 9,000 problems to measure the practical capabilities of AI, it ranked fifth out of around 500 models, surpassing those from Google and others.
Many US AI experts have likened it to China's <a href="/company/deepseek">DeepSeek</a>, whose R1 model appeared in January 2025 and stunned the world with its capabilities and low costs.
Since the 2022 public release of OpenAI's ChatGPT, American companies have led the way in AI performance, with Chinese companies following closely behind. Most US AI models are closed, meaning that their technology is kept secret, while many Chinese companies build inexpensive and easily replicable open-source models.
As AI evolves, soaring usage fees have emerged as a hurdle. More complex tasks lead to the use of more tokens, pushing up costs. Some major US companies, including Airbnb and Uber Technologies, have acknowledged internal use of Chinese AI.
US President Donald Trump's administration, previously reluctant to regulate AI, imposed export controls on Anthropic's Claude Mythos 5 and Claude Fable 5 in June, resulting in the company temporarily suspending both models.
The administration's shift to a more controlling stance dramatically increased the risk of sudden action making services unavailable, leading more companies to adopt Chinese models.
Data from US startup OpenRouter—which offers a service used by more 8 million users, primarily engineers, that allows them to switch freely between more than 400 types of AI—was used to compare the number of tokens consumed by US and Chinese AI.
Chinese models, which had less than half the number of tokens used as US models at the start of the year, surpassed US models multiple times after February. Their use exploded in June, reaching 25 trillion tokens in the final week of the month, more than double the number at the end of May and 78% more than US models.
DeepSeek held the top spot in market share by company in June, at 19%. Such other Chinese companies as Z.ai and <a href="/company/xiaomi">Xiaomi</a> also saw significant growth, while Google's and OpenAI's shares shrank.
"With cost effective token usage dominating the summer headlines, we'd expect these cheaper-but-still-powerful [Chinese] open source models to continue to attract more tokens over time," OpenRouter said in a report.
The US is increasingly wary of the rapid advances in Chinese AI. Anthropic CEO Dario Amodei has posited that AI will be used to strengthen authoritarian states like China, including through surveillance of citizens, leading to a decline in democracy.
A potentially greater threat is Chinese AI companies engaging in distillation, using the output of high-performance AI models to develop other AI, reducing costs while maintaining performance. Many US companies prohibit unauthorized distillation.
In mid-June, Anthropic wrote to US senators to <a title="" href="https://asia.nikkei.com/business/technology/artificial-intelligence/anthropic-accuses-alibaba-of-largest-known-distillation-attack-on-claude" target="_blank" rel="noopener">demand action against Alibaba</a>, accusing the Chinese technology group of "the largest known distillation attack" on the company to date.
If distillation becomes persistent, even if US companies continue to invest massive funds to develop cutting-edge AI, Chinese companies will be able to quickly catch up.
Using open-source AI models comes with its own risks, as there is no guarantee that Chinese companies will continue to provide their models to external parties indefinitely. There is also a chance that high-performance Chinese AI comparable to Claude Mythos could be developed.
"Open-source models are difficult to regulate with suspensions or prescreening measures because they can be replicated," said Boston Consulting Group's Shinichi Takayanagi, an expert on AI. "If they are available at a low price, it will put pressure on US AI companies to lower prices too, resulting in insufficient investments and potentially leading to neglect of AI safety measures."
This article first appeared on <a href="https://asia.nikkei.com/business/technology/artificial-intelligence/chinese-ai-usage-by-us-firms-soared-after-mythos-restrictions">Nikkei Asia</a>. It has been republished here as part of 36Kr’s ongoing partnership with <a href="https://www.0722vps.aieehr01.top/nikkei-teams-up-with-36kr-in-asia-tech-news-coverage/">Nikkei</a>.

Chinese AI usage by US firms soared after Mythos restrictions